Artists are caught in the crosshair. A world of “we will never get automated” just got automated. In the past few years artwork is now AI Assisted; with simple, open source, and wildly spread code. The world of AI Assisted artwork is a blossoming flower to most people, however there’s a world of theft and ethical practices being consistently misunderstood. Thought leaders are created over night without any identification possible. Will the government ever catchup?
My theory; By creating a system to take advantage of open source models/algorithms that can steal from other artists without their knowledge, I can increase the visibility into unethical practices, and demonstrate that anyone can stylize an image. I grabbed a tutorial from tensorflow, their easiest tutorial, and turned it into a “LinkedIn profile” image input, plus someone’s street artwork, equals my net new image.
My goal; Improve AI assisted artwork ethical practices by demonstrating how unethical it can be. Give end users the ability to see where the problems live by expressing it in text below.
Hello and welcome to this Collab python book (links take you to google free python application stored in their cloud, where I’ve tested this application many times, and it contains my opinions in this top half and below is a complete set of code that will allow you to quickly stylize images of your choosing. If you want to build this environment on your computer, check out my blog on setting up anaconda distribution – a data science toolkit.
What is this tldr content by tyler garrett? I’m talking about fast style transfer for arbitrary styles with beginner notes to help anyone use the solution here with “url” links to two images. You can use your LinkedIn photo, or any image online (most image URLs will work). I leave an image of myself (LinkedIn) and an image of a planet for example material.
We are seeing a more attention focused in AI assisted artwork and I feel obliged to offer some advice/training. I think it’s necessary to say that I feel this is unethical if used incorrectly and at the same time it’s incredible cool because I’ll never in my lifetime create something this mesmerizing in this amount of time.
It is becoming easier than ever to take photos of some ones painting in real life or street artwork, and use it as a style for my profile image on LinkedIn. Below this essay are 4 different photos/examples.
Unethical practices; This is where the gray area presents itself, what if I’m taking images from an ecommerce website and using it to create my own artwork. Their artwork as an input. What if I profit from this artwork?
Where do we draw the line and will anyone notice the differences? When automating style transfers into our content, are we considering ethics or are we throwing it out the door?
I believe we are always building on someone else’s artwork, and there’s nothing truly original or unique. However if I can grab some ones ecommerce store product and turn it into my own art, do you feel I’ve cross a gray area?
The future will be full of scam artists; Creating artwork automatically while pretending the AI didn’t assist, now you’re thought leader-ing you’re a great artist… well this might be the next level of fame, or are we already there today? (We are already seeing it today on social media)
A lot of accounts online today are not images of humans, fake accounts, or what they call “sock puppet accounts.” Who’s to say some of the famous ones aren’t using AI to generate their fame, revenue, and the more you think about it.. Doesn’t sound half bad for a computer to do your work.
If NFT artists using AI artwork as their entire recipe for success, do we care, or are we trying to earn money? Will this small chunk of code be successful at creating money? I don’t know atm, but currently the NFT industry is down 50%.
Questions to ask before buying NFT or using these AI apps.
Numbered below however the moral of the story is “who am I stealing from” and “should I care?”
- Should the original artist be compensated or at a bare minimum notified?
- Will technology catch up to the point of seeing who’s using someone else’s work to create their own or maybe a reverse engineering tool?
- Is that even possible?
- Also, should we start asking ourselves if the artwork we are seeing today is either created by an artist or a machine…
- Does it matter?
It would seem important that we manage this because it’s going to get out of hand but how exactly will we do that? I’ve read plenty of articles about the negative side effects, however the writers aren’t using the technology most of the time and that means they have no feedback on the capability or even flexibility. When you put it that way, maybe we should start consider starting to govern these AI applications prior to generation? Maybe a best practice for AI for ethics reasonings?
Another oddity is people are now asking AI to draw for them. A strange idea to think we would ask a computer to do a hand drawing but that’s what’s happening and that’s starting to scare a lot of artists because they are starting to look more and more realistic. Also, you can say “hand drawn with pencil” and the AI will make it seem like someone actually did this manually..
The dark side; Do you want someone training their art model using images of your children and what would be the damage to you if they did? This is enough to keep me from posting any photos of my children on the internet.
Hello, my name is Tyler Garrett and I’ve been in the enterprise technology solutions realm and business intelligence for the past 12 years. I’m excited to make this ‘AI assisted’ stylization artwork blog because if we make this more accessible for non-technical end users, we will gain more visibility into the ethics concerns that will happen in Ai assisted art world.
Dark Side 2.0; NFTs could be artists stealing someone else’s work and profiting anonymously. There’s no real governance.
The dark side of NFT is anyone can profit infinitely on anyone else’s work. I was interested in this realm because I had an idea that was relevant to stylizing videos for music, however when I got started demoing this application in google collaboratory, I found I was stealing peoples street art and they have no idea.
Then we have open sourced AI assisted artwork being created from text, which is starting to bridge deeper ethics boundaries related to “permission and consent.” People are not giving permission to others, like celebrities are not approving nude images to be created, which is leading down a very devious concern for many parents, celebrities, and people who have the time to plug in and understand what’s happening.
Because of the boom, it feels like an appropriate time to be diligent with the documentation and type more information here than engineers would normally write. The more visibility we can bring to the topic from non-technical people, the better the adoption and chance of improved future algorithms.
Computer vision technology will expand because smart people like to share the things they are developing and even if it can be an ethics concerns they are not thinking about this rather they are often thinking of growing their following by being a creator of something they believe is technical and cool. Not all creators are like this but most enjoy the attention. I don’t mean to be rude, I too fall victim to the gamification of social media platforms and would love to be recognized for creating cool tech.
Engineering at these high levels means you normally don’t explain much and expect everyone to understand or read documentation written in the same difficult method.
I feel it’s important someone simplifies the logic for people who are not engineers because we need the feedback from people who are not engineers too. The faster everyone catches up to the small percent of people who understand computer vision, the faster the laws will catch up, and this will increase adoption of the usage of AI assisted artwork in a positive way (i hope) OR the removal of bad actors, quickly.
Engineers who understand this technology are very rare and likely .001% of the population or less. That means a very small group of people will be able to suggest the ethics are right or wrong. We need more visibility into this topics and thus I’ve created this Fast Style Transfer for Arbitrary Styles for Beginners & notes from Tyler Garrett.